Volumetric Data Segmentation via Skeleton Graph Branching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for segmenting and classifying voxels in medical images, such as CT images, are tedious and time-consuming, requiring manual review and marking of pixels, which hinders efficient surgical or treatment planning.
Innovation Solution
A method involving CT image data processing that includes segmentation, skeletonization, graphing, and 3D mesh model generation, where each voxel is associated with a branch ID, enabling automated identification and visualization of structures like airways and blood vessels, and allowing for user interface highlighting and clustering for improved analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual segmentation and classification of voxels is performed, then measurement precision and reliability of anatomical structure identification is improved, but loss of time and productivity deteriorates significantly
Solution Approach 1:
The system performs automated segmentation and classification of voxels without requiring manual intervention. The algorithm independently processes CT image data, identifies anatomical structures, and generates 3D models automatically, eliminating the need for clinicians to manually review and mark pixels while maintaining accurate results
Solution Approach 2:
The patent replaces the mechanical manual process of pixel-by-pixel marking with an automated computational algorithm. The system uses image processing techniques, skeletonization, graph theory, and mesh generation to automatically segment and classify voxels, substituting human manual operations with automated digital processing
2Productivity
If automated segmentation methods are used, then productivity and speed of analysis is improved, but device complexity and difficulty of detecting and measuring increases
Solution Approach 1:
The system divides the complex task of volumetric segmentation into distinct sequential steps: initial segmentation of voxels, skeletonization to identify center points, graphing to create branching structures, and mesh generation to produce 3D models. This segmentation of the processing pipeline manages complexity through modular, manageable stages
Solution Approach 2:
The patent introduces intermediate representations to bridge the gap between raw CT data and final 3D models. Skeletonization creates a simplified center-point representation, graphing converts this to a branching structure with nodes and edges, and mesh generation finally produces the 3D surface. These intermediaries simplify the overall complexity by breaking down the transformation process
3Measurement precision
If manual pixel marking is performed across multiple 2D slices, then measurement precision is improved, but ease of operation deteriorates due to tedious scrolling and marking processes
Solution Approach 1:
The automated system performs segmentation and classification independently without requiring operator intervention. The algorithm automatically processes each slice, identifies structures, and builds the 3D model without needing the clinician to scroll through slices or mark pixels, dramatically improving ease of operation
Solution Approach 2:
The system performs preliminary automated segmentation and classification of all voxels before any user review. The skeletonization, graphing, and mesh generation are completed automatically in advance, so that when clinicians view the results, the complex processing is already done, eliminating the need for manual slice-by-slice review
Data Source
AI summary
A system and method of image processing including a processor in communication with a display and a computer readable recording medium having instructions executed by the processor to read an image data set from the memory, segment the image data set, and skeletonize the segmented image data set. The instructions cause the processor to graph the skeletonized image data set, assign a branch identification (ID) for each branch in the graph, and associate each voxel of the segmented image data set with a branch ID. The instructions cause the processor to generate a three-dimensional (3D) mesh model from the graphed skeletonized image data set, associate each vertex of the 3D mesh model with a branch ID; and display in a user interface the 3D mesh model, and an image of the image data set.